index.html

<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="UTF-8" />
  <meta name="viewport" content="width=device-width, initial-scale=1.0" />
  <title>Bingchen Huang</title>
  <meta name="description" content="Bingchen Huang (黄丙晨) — Agent Algorithm Engineer at Meituan. AutoResearch, self-evolving agents, and reliable long-horizon research." />
  <meta name="author" content="Bingchen Huang" />

  <!-- Open Graph -->
  <meta property="og:type" content="website" />
  <meta property="og:title" content="Bingchen Huang" />
  <meta property="og:description" content="Agent Algorithm Engineer at Meituan. AutoResearch &amp; Self-Evolving Agents." />
  <meta property="og:url" content="https://yellowpancake.github.io/" />

  <link rel="icon" type="image/png" href="images/site-logo.png" />
  <link rel="preconnect" href="https://fonts.googleapis.com" />
  <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />
  <link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&family=Newsreader:ital,opsz,wght@0,6..72,400;0,6..72,500;0,6..72,600;1,6..72,400&display=swap" rel="stylesheet" />
  <link rel="stylesheet" href="assets/css/main.css" />
</head>
<body>
  <!-- ===================== NAV ===================== -->
  <header class="nav" id="nav">
    <div class="nav__inner">
      <a class="nav__brand" href="#top">BH</a>
      <nav class="nav__links">
        <a href="#about">About</a>
        <a href="#news">News</a>
        <a href="#research">Research</a>
        <a href="#publications">Publications</a>
        <a href="files/BingchenHuang.pdf" target="_blank" rel="noopener">CV <span lang="zh-CN">中文</span></a>
      </nav>
      <button class="theme-toggle" id="themeToggle" aria-label="Toggle theme" title="Toggle dark / light">
        <span class="theme-toggle__icon">☾</span>
      </button>
    </div>
  </header>

  <main id="top">
    <!-- ===================== HERO / ABOUT ===================== -->
    <section class="hero" id="about">
      <div class="hero__grid">
        <div class="hero__text">
          <p class="hero__eyebrow">AutoResearch &amp; Self-Evolving Agents</p>
          <h1 class="hero__name">Bingchen Huang <span class="hero__cn">黄丙晨</span></h1>
          <p class="hero__role">Agent Algorithm Engineer · <a href="https://www.meituan.com/" target="_blank" rel="noopener">Meituan</a></p>

          <p class="hero__bio">
            I focus on <strong>AutoResearch</strong> and <strong>self-evolving agents</strong>:
            building agents that carry out research and improve through experience.
            I initiated and lead <a href="#dr-pie">Dr. Pie</a>, an enterprise research agent
            combining research skills, execution harnesses, and skill evolution.
          </p>
          <p class="hero__bio">
            My earlier work focused on large-model post-training, multimodal in-context learning,
            and continual learning. I received my M.Sc. from the
            <a href="https://fvl.fudan.edu.cn/" target="_blank" rel="noopener">FVL Lab</a>,
            <a href="https://www.fudan.edu.cn/" target="_blank" rel="noopener">Fudan University</a>,
            advised by Prof. Yu-Gang Jiang and Prof. Zhineng Chen, and my B.Eng. from
            <a href="https://www.ecust.edu.cn/" target="_blank" rel="noopener">East China University of Science and Technology</a>.
          </p>

          <ul class="hero__meta">
            <li><span>📍</span> Beijing, China</li>
            <li><span>🏢</span> Meituan</li>
            <li><span>✉️</span> <a href="mailto:huangbchen@foxmail.com">huangbchen@foxmail.com</a></li>
          </ul>

          <div class="hero__links">
            <a class="btn btn--primary" href="https://scholar.google.com/citations?user=nX3Kp9oAAAAJ&hl=en" target="_blank" rel="noopener">Google Scholar</a>
            <a class="btn" href="https://github.com/YellowPancake" target="_blank" rel="noopener">GitHub</a>
            <a class="btn" href="https://www.linkedin.com/in/bc-huang" target="_blank" rel="noopener">LinkedIn</a>
            <a class="btn" href="files/BingchenHuang.pdf" target="_blank" rel="noopener">CV · 中文 PDF</a>
          </div>
        </div>

        <div class="hero__photo">
          <img src="assets/img/avatar.jpg" alt="Portrait of Bingchen Huang" loading="eager" />
        </div>
      </div>
    </section>

    <!-- ===================== NEWS ===================== -->
    <section class="section reveal" id="news">
      <h2 class="section__title"><span class="section__num">§ 01</span> News</h2>
      <ul class="news">
        <li><time datetime="2026-07-02">2026.07</time><p>Our paper <strong>DRDN</strong> is released on <a href="https://arxiv.org/abs/2607.01630" target="_blank" rel="noopener">arXiv</a>.</p></li>
        <li><time datetime="2026-07-01">2026.07</time><p>Our paper <strong>ReVisIT</strong> is released on <a href="https://arxiv.org/abs/2607.00606" target="_blank" rel="noopener">arXiv</a>.</p></li>
        <li><time datetime="2026-02">2026.02</time><p>Started <strong>Dr. Pie</strong>, an enterprise AutoResearch agent, as project initiator and lead.</p></li>
        <li><time>2025.07</time><p>Our paper <strong>TDR</strong> (Task-Decoupled Retrieval for In-Context Learning) is released on <a href="https://arxiv.org/abs/2507.18340" target="_blank" rel="noopener">arXiv</a>.</p></li>
        <li><time>2024.04</time><p>Contributed to the <strong>NTIRE 2024</strong> Quality Assessment of AI-Generated Content Challenge (CVPR 2024 Workshops).</p></li>
        <li><time>2023.07</time><p><strong>MRN</strong> is accepted to <strong>ICCV 2023</strong>.</p></li>
        <li><time>2022.11</time><p><strong>TCIL</strong> is accepted to <strong>AAAI 2023</strong>.</p></li>
      </ul>
    </section>

    <!-- ===================== RESEARCH ===================== -->
    <section class="section reveal" id="research">
      <h2 class="section__title"><span class="section__num">§ 02</span> Research Interests</h2>
      <div class="cards">
        <article class="card">
          <h3>AutoResearch</h3>
          <p>Composable research skills and agents that connect problem formulation, experimentation, evaluation, and scientific writing.</p>
        </article>
        <article class="card">
          <h3>Self-Evolving Agents</h3>
          <p>Learning from execution traces and failures to improve skills through reflection, offline evolution, and task-level validation.</p>
        </article>
        <article class="card">
          <h3>Long-Horizon Reliability</h3>
          <p>Execution harnesses, persistent state, process verification, and recovery for research workflows that run over days.</p>
        </article>
      </div>
      <article class="project" id="dr-pie">
        <div class="project__heading">
          <div>
            <p class="project__eyebrow">Current project · Initiator &amp; Lead</p>
            <h3>Dr. Pie <span>Enterprise AutoResearch Agent</span></h3>
          </div>
          <span class="badge">2026 — Present</span>
        </div>
        <p>I initiated Dr. Pie and built its early research skills, core harness, self-evolution mechanisms, and desktop application. I now lead its technical direction, development, and benchmark evaluation.</p>
        <p>The system combines composable research workflows with state management and recovery for long-running tasks. It learns from execution experience through offline skill evolution, with validation before adopting changes.</p>
        <p class="project__footnote">Used internally at Meituan by over a thousand users, including in company-level model research workflows.</p>
      </article>
      <p class="research__background"><strong>Earlier work:</strong> large-model post-training (SFT / RL), multimodal in-context learning, domain embeddings, and continual learning.</p>
    </section>

    <!-- ===================== PUBLICATIONS ===================== -->
    <section class="section reveal" id="publications">
      <h2 class="section__title"><span class="section__num">§ 03</span> Publications</h2>
      <p class="section__hint"><span class="hl">*</span> Equal contribution (co-first authors). Published papers and preprints; ongoing work is listed separately below. See <a href="https://scholar.google.com/citations?user=nX3Kp9oAAAAJ&hl=en" target="_blank" rel="noopener">Google Scholar</a> for the full list.</p>

      <ol class="pub">
        <li class="pub__item">
          <div class="pub__venue"><span class="badge badge--prep">arXiv 2026</span></div>
          <div class="pub__body">
            <h3 class="pub__title">Retrieved Images as Visual Thought: Training-Free Multimodal In-Context Learning for the Open-vs-Closed Gap</h3>
            <p class="pub__authors"><span class="me">Bingchen Huang</span>, Zhiling Wang, Yifu Chen, Yuanchao Du</p>
            <p class="pub__sub">ReVisIT · Retrieved image exemplars as visual thought for multimodal in-context learning.</p>
            <p class="pub__links">
              <a href="https://arxiv.org/abs/2607.00606" target="_blank" rel="noopener">arXiv</a>
              <a href="https://arxiv.org/pdf/2607.00606" target="_blank" rel="noopener">PDF</a>
            </p>
          </div>
        </li>
        <li class="pub__item">
          <div class="pub__venue"><span class="badge badge--prep">arXiv 2026</span></div>
          <div class="pub__body">
            <h3 class="pub__title">DRDN: Decoupled Representation Dynamic Network for From-Scratch ViT Class-Incremental Learning</h3>
            <p class="pub__authors"><span class="me">Bingchen Huang</span>, Yifu Chen, Zhiling Wang, Yuanchao Du</p>
            <p class="pub__sub">Decoupling shared representations and task-specific knowledge for continual learning.</p>
            <p class="pub__links">
              <a href="https://arxiv.org/abs/2607.01630" target="_blank" rel="noopener">arXiv</a>
              <a href="https://arxiv.org/pdf/2607.01630" target="_blank" rel="noopener">PDF</a>
            </p>
          </div>
        </li>

        <li class="pub__item">
          <div class="pub__venue"><span class="badge badge--prep">arXiv 2025</span></div>
          <div class="pub__body">
            <h3 class="pub__title">TDR: Task-Decoupled Retrieval with Fine-Grained LLM Feedback for In-Context Learning</h3>
            <p class="pub__authors">Yifu Chen<sup>*</sup>, <span class="me">Bingchen Huang<sup>*</sup></span>, Zhiling Wang, Yuanchao Du, Junfeng Luo, Lei Shen, Zhineng Chen</p>
            <p class="pub__links">
              <a href="https://arxiv.org/abs/2507.18340" target="_blank" rel="noopener">arXiv</a>
            </p>
          </div>
        </li>

        <li class="pub__item">
          <div class="pub__venue"><span class="badge">CVPR-W 2024</span></div>
          <div class="pub__body">
            <h3 class="pub__title">NTIRE 2024 Quality Assessment of AI-Generated Content Challenge</h3>
            <p class="pub__authors">Xiaohong Liu, Xiongkuo Min, Guangtao Zhai, et al. (incl. <span class="me">Bingchen Huang</span>)</p>
            <p class="pub__sub">IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops</p>
            <p class="pub__links">
              <a href="https://arxiv.org/abs/2404.16687" target="_blank" rel="noopener">arXiv</a>
            </p>
          </div>
        </li>

        <li class="pub__item">
          <div class="pub__venue"><span class="badge badge--top">ICCV 2023</span></div>
          <div class="pub__body">
            <h3 class="pub__title">MRN: Multiplexed Routing Network for Incremental Multilingual Text Recognition</h3>
            <p class="pub__authors">Tianlun Zheng, Zhineng Chen, <span class="me">Bingchen Huang</span>, Wei Zhang, Yu-Gang Jiang</p>
            <p class="pub__sub">IEEE/CVF International Conference on Computer Vision</p>
            <p class="pub__links">
              <a href="https://arxiv.org/abs/2305.14758" target="_blank" rel="noopener">arXiv</a>
              <a href="https://openaccess.thecvf.com/content/ICCV2023/papers/Zheng_MRN_Multiplexed_Routing_Network_for_Incremental_Multilingual_Text_Recognition_ICCV_2023_paper.pdf" target="_blank" rel="noopener">PDF</a>
              <a href="https://github.com/YellowPancake/MRN" target="_blank" rel="noopener">Code</a>
            </p>
          </div>
        </li>

        <li class="pub__item">
          <div class="pub__venue"><span class="badge badge--top">AAAI 2023</span></div>
          <div class="pub__body">
            <h3 class="pub__title">Resolving Task Confusion in Dynamic Expansion Architectures for Class Incremental Learning</h3>
            <p class="pub__authors"><span class="me">Bingchen Huang</span>, Zhineng Chen, Peng Zhou, Jiayin Chen, Zuxuan Wu</p>
            <p class="pub__sub">Proceedings of the AAAI Conference on Artificial Intelligence</p>
            <p class="pub__links">
              <a href="https://arxiv.org/abs/2212.14284" target="_blank" rel="noopener">arXiv</a>
              <a href="https://ojs.aaai.org/index.php/AAAI/article/view/25170" target="_blank" rel="noopener">AAAI</a>
              <a href="https://github.com/YellowPancake/TCIL" target="_blank" rel="noopener">Code</a>
            </p>
          </div>
        </li>
      </ol>
      <div class="ongoing">
        <h3 class="ongoing__label">Ongoing Research</h3>
        <article>
          <p class="project__eyebrow">Agent self-evolution · Manuscript in preparation · 2026</p>
          <h4 class="pub__title">GRSO: Group-Relative Skill Optimization for Frozen Agents</h4>
          <p class="ongoing__description">Optimizing agent skills with problem-level group sampling, contrastive reflection on successful and failed trajectories, and per-problem validation, while keeping model weights frozen.</p>
        </article>
      </div>
    </section>
  </main>

  <!-- ===================== FOOTER ===================== -->
  <footer class="footer">
    <p>© <span id="year">2026</span> Bingchen Huang · Built with plain HTML &amp; CSS on GitHub Pages.</p>
    <p class="footer__sub">Design inspired by clean academic homepages. Last updated September 2026.</p>
  </footer>

  <script src="assets/js/main.js"></script>
</body>
</html>